Re: bug report:

[Kevin Ryde]

Jean-Christophe Filliatre <address@hidden> writes:
> 
> I think  I've found a  bug of mpz_fdiv_q.  The file bug.c  is attached
> below. Here is the bug report:
> 
> ======================================================================
> GMP version number: 3.1.1
> 
> test program: bug.c
> 
> compilation: gcc -c -I. bug.c .libs/libgmp.a
> 
> what is wrong: execution crashes on a floating point exception

Is this a divide by zero?  You could put it under gdb and do a stack
backtrace with "where" if you like.  You could check "b" isn't zero
for some strange reason too.

If the backtrace is uninformative you could try a build with just
CFLAGS=-g to avoid -fomit-frame-pointer (and hope very hard the
problem still arises).


> Configuration options:
>    ./configure --prefix=/usr
> 
> gcc -v:
>    Reading specs from /usr/lib/gcc-lib/i686-pc-linux-gnu/pgcc-2.91.66/specs
>    gcc version pgcc-2.91.66 19990314 (egcs-1.1.2 release)
> 
> uname -a: 
>    Linux gemini 2.2.13-7mdk #1 Wed Sep 15 18:02:18 CEST 1999 i686 unknown
>  
> ./config.guess:
>    pentium2-pc-linux-gnu

I don't think I've got that compiler handy.  Is it the pentium hacked
gcc from a while ago (or maybe I'm thinking of something else)?


> ======================================================================
> 
> I noticed that bug doesn't  reproduce on another of my machines, where
> we have:
> 
> ======================================================================
> gcc -v:
>    Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/egcs-2.91.66/specs
>    gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)
> 
> uname -a:
>    Linux pc803 2.2.17 #2 Thu Sep 21 16:18:38 MEST 2000 i686 unknown
> 
> ./config.guess:
>    pentium3-pc-linux-gnu
> ======================================================================
> 
> So  I guess it  is related  to specialized  assembly code  for Celeron
> processors.

The differences between pentium2 and pentium3 are minimal, so it's
perhaps more likely to be the compiler.


> (This bug is  quite annoying for me, so I would  appreciate a patch if
> you can provide one easily.)

You can build on p3 for p2,

        ./configure --target=pentium2-pc-linux-gnu

The only differences currently are the mpn_popcount and mpn_hamdist
implementations, so in fact for the test program the p3 build should
run already.